水下智能潜器的神经网络运动控制

本文介绍一种基于神经网络的水下智能潜器的运动控制方法，该方法通过在线学习，融控制与滤波为一体。计算机仿真与水池实验验证表明，该方法的控制与滤波性能良好，对环境的学习与适应能力强。该方法事实上可用于一般动力系统的控制。     

PLS-ANN分析血浆自体荧光光谱的二阶导数光谱识别胃癌

探讨偏最小二乘法结合神经网络法(简称PLS-ANN)分析血浆自体荧光光谱的二阶导数光谱识别胃癌的优势,对20例胃癌病人和23例健康人血浆进行以405nm为激发光的自体荧光光谱检测,采用PLS-ANN法分别对血浆自体荧光光谱和二阶导数光谱进行判别分析。PLS-ANN分析血浆的自体荧光光谱法诊断胃癌的灵敏度为75%,特异度为83%,准确率为79%,PLS-ANN分析血浆的二阶导数光谱法诊断胃癌的灵敏度为90%,特异度为96%,准确率为93%。结果表明PLS-ANN分析血浆的二阶导数光谱法识别胃癌,优于PLS-ANN分析血浆的自体荧光光谱法,有望成为快速识别胃癌的较理想方法。     

Resistant performance of perforation of multi-layered targets using an estimation procedure with marine application

Recently, multi-layered targets have become commonplace in both military and civilian applications, such as marine hulls, armored vehicle bodies, outside structures of bulletproof cars, and aerospace vessels. This paper studies the resistant performance of perforation in multi-layered targets. An estimation procedure is established based on the concepts of the conservation of momentum, impulse-momentum law, and conservation of energy. Experimental results of the test of Almohandes et al. are adopted to check the residual velocity of multi-layered targets. The results of the verification are good in terms of agreement for impact velocities ranging from 700 to 800 m/s, when the ratio of the projectile length to the projectile diameter (i.e. L/D) is 4.2, and the average residual velocity error of single, double- or triple-layered targets range approximately from 4.42 to 8.40%. The ballistic performance is best for the double target when the ratio of the first layer thickness to the total thickness (i.e. t₁/(t₁+t₂)) is about 0.75, and the worse performance occurs when the ratio t₁/(t₁+t₂) is 0.5. An air gap slightly influences the resistant performance of perforation in multi-layered targets. These results may serve as a useful reference for designers.     

大民屯凹陷地层水水文地质特征与油气聚集关系

对大民屯凹陷前第三系潜山、下第三系沙河街组地层水化学场研究发现，本区油田水总矿化度及二价阳离子浓度偏低，地层水总矿化度具有明显的垂向分带性，水体主要为陆相成因并受地表水渗入的影响较大，变质程度高，油田水的演化与盐类的溶解有关。该区地层水的流动受断裂和沉积砂体的控制，在水头压力、浮力动力场控制下其流动样式和分布规律反映了凹陷内油气的运聚规律，可为本区自生自储、新生古储式油气藏的判别提供有利证据。     

基于ArcEngine的数字高程模型更新方法研究

采用COM技术,以ARCENGINE所提供的类库和凸包算法,实现了数字高程模型(DEM)数据更新的边界检测和利用矢量地形数据采集更新数字高程模型数据的方法。保证了数字高程模型数据的精度,可以满足数字高程模型的现势性需求。     

曲流点坝薄夹层构形对驱油效率及剩余油形成与分布的影响

陆相储层单砂体内的薄夹层是形成储层流体流动非均质的主要因素之一,单砂体内部薄夹层极薄、规模小、测井全部识别有一定困难,且位于层内。在油田开发初、中期作用不明显,而在高含水期、三采阶段,单砂体及其渗透率宏观分布已不能详尽揭示层内剩余油分布,层内薄夹层对注入剂驱油的影响及其重要性逐渐显现出来,与渗透率在储层内韵律性分布有同等重要的地位,也是储层精细表征的重要内容。以往只是着重从渗透率在储层内分布的不均衡的角度来研究储层中驱油效率问题,忽略了薄夹层的对储层驱油效率的影响。以大庆油田葡萄花油层组PⅠ2小层曲流河道砂体为例,对曲流河道砂体内部薄夹层构形对驱油效率及剩余油的形成与分布作了初步的分析和探讨。在排除开发因素差异的同井单砂体分析条件下,提出了“单砂体内部薄夹层空间构形 渗透率垂向序列 重力”三因素共同控制剩余油分布、驱油效率多段垂向序列模式。     

Seasonal variability of intrusion of the Kuroshio water across the continental shelf northeast of Taiwan

The currents over the East China Sea continental slope northeast of Taiwan were monitored from mid-July to early December, 1992 and again from late March to early August, 1993, covering the two transition periods of monsoon seasons. In 1992, we found massive Kuroshio intrusion (onshore velocity reaching 25 cm/s or more) occurred at mid-October, a lag of one month after the wind pattern changed from southerly to northeasterly. By comparing the surface meteorological with observed cross-slope velocity, we suggest that cooling rather than direct wind forcing, i.e., mid-October cold air outbreaks associated with the Siberia highs moving across the East China Sea, may be the major cause for triggering the observed winter intrusion event. The intrusion continued until the end of the 1992 data in early December and again in late March and early April of 1993 when the data became available. In mid-April, 1993, the intrusion suddenly halted when the lows replaced the highs and the wind reversed with warm air from south. In addition, we found that short-period intrusions appeared in August and early September, 1992 and in late July, 1993. All of these summer intrusions coincide with the occurrence of typhoon passages over or further offshore of the area northeast of Taiwan.     

应用遗传神经网格方法分析赤潮监测数据

简要介绍了遗传神经网络方法的原理 ,探讨了应用遗传神经网络方法研究辽东湾海域丹麦细柱藻 (Leptocylindrusdanicus)赤潮与其环境因子之间的关系 ,计算了各环境因子对丹麦细柱藻赤潮的贡献。结果表明 ,温度、盐度、DIN的变化对研究海域丹麦细柱藻种群密度的增长有比较重要的影响 ,DIN是营养限制因子。遗传神经网络是分析赤潮监测数据的有效方法 ,值得进一步探索     

琉球群岛东部海区表层沉积物中浮游有孔虫分布及指示意义

对琉球群岛东部海区123个表层沉积物样品中浮游有孔虫因子分析结果表明,浮游有孔虫组合在溶跃面以上表现为Globigerinoides ruber-Globigerinata glutinata组合,溶跃面以下为Globorotalia inflata-Neoglobo-quadrina pachyderma (dex.)组合,局部受黑潮影响区域黑潮标志种Pulleniatina obliquiloculata含量相对丰富,为Neogloboquadrina dutertrei-P.obliquiloculata组合;表层沉积物中浮游有孔虫的丰度及组合分布特征对于碳酸盐溶跃面深度及碳酸盐补偿深度有一定的指示意义。另外,碳酸盐保存特征研究表明,该区域现代碳酸盐溶跃面深度约为3100 m,碳酸盐补偿深度约为4300 m。     

基于GIS的海洋环境质量评价及其可视化研究

利用GIS基础平台软件的空间数据拓扑操作优势,结合海洋环境质量评价数学模型,实现了海洋环境质量评价过程及结果表达的可视化。提高对海洋环境质量评价结果的科学性、现势性和直观化,为领导的决策提供有力的支持。